Exciting news! Today we are releasing Cerebras-GPT, a family of 7 GPT models from 111M to 13B parameters trained using the Chinchilla formula. These are the highest accuracy models for a compute budget and are available today open-source! (1/5) Press:
@cerebras
-
CEO speaks at University of Chicago Caltech AI Science Conference
By
–
Excited to have our CEO, Andrew Feldman, participate as a panelist speaker at the University of Chicago and Caltech Conference on AI+Science. We appreciate being recognized for building the future of AI-enabled scientific discoveries More info here – https://
hubs.li/Q01J69TL0 -
Sparsity Improves AI Model Accuracy Through Sparse Transformations
By
–
This is the first work (that we know of!) to demonstrate the use of #sparsity for improving the accuracy of models via a set of sparse transformations. If you're interested in learning more about SIFT, check out our blog – https://
cerebras.net/blog/can-spars
ity-make-ai-models-more-accurate
… (4/4) -
ResNet and GPT-3 achieve efficiency gains matching larger models
By
–
Some of the highlights of this work include ResNet-18 on ImageNet achieving a 3.5% accuracy improvement, and #GPT-3 Small on WikiText-103 reducing perplexity by 0.4, both matching larger dense model variants that have 2x or more FLOPs. (3/4)
-
Cerebras Enables 50k Context Windows for AI Models
By
–
Leveraging context windows up to 32k is an accomplishment that only a handful of companies can enable. At Cerebras, we empower everyone from beginners to experts with the ability to leverage context windows up to 50k with just a few clicks! Learn more :
-
Cerebras LLMs for Healthcare Document Summarization
By
–
Join Cerebras Director of Product Management Natalia Vassilieva at the Healthcare NLP Summit and learn how large language models are used for long document abstractive summarization.
-
Sparsity Accelerates GPT Training While Preserving Model Accuracy
By
–
Sparsity can improve your model's training performance! Our research shows the promise of high sparsity on large-scale GPT models, providing training acceleration at a fraction of the FLOPs while preserving downstream accuracy. Learn more here –
https://
cerebras.net/blog/accelerat
ing-large-gpt-training-with-sparse-pre-training-and-dense-fine-tuning/?utm_content=241655076&utm_medium=social&utm_source=twitter&hss_channel=tw-751545566778171392
… -
OpenXLA Expands Cerebras Wafer-Scale Engine ML Framework Access
By
–
“OpenXLA helps extend our user reach and accelerated time to solution by providing the Cerebras Wafer-Scale Engine with a common interface to higher level ML frameworks,” says our VP of Product, Andy Hock Learn more about the OpenXLA Project here:
-
Cerebras Advances AI Accessibility Through OpenXLA Framework
By
–
Cerebras builds leading AI systems and software to make large-scale AI easy and accessible to organizations. We are proud to see OpenXLA, a solution that makes frameworks easier to deploy across different hardware options, being made generally available and accessible to all.
-
Cerebras Systems Named Forbes Best Startup Employer
By
–
Cerebras Systems has been recognized by Forbes as one of America's Best Startup Employers! https://
hubs.li/Q01FSHlp0